Research And Empirical Analysis Of Statistics On Arbitrage Strategy Of Commodity Futures Under High Frequency Data

As a more sophisticated quantitative trading strategy,statistical arbitrage strategies have long been used in the investment in developed capital markets.Compared to overseas mature markets,domestic quantitative investment is still in its infancy and the overall size of quantitative investment products is still small.Quantitative investment strategies are relatively single and lack the support of multiple quantitative strategies.And because domestic stock markets and futures markets are relatively immature,and they are often in volatile fluctuations,it is extremely necessary to study the strategy of controlling risk and stable income,while statistical arbitrage strategies are more stable than unilateral trading strategies and bring fluctuations to funds.Not too big,so it is necessary to carry out in-depth and extensive research on quantitative investment statistical arbitrage.The main content,research methods and conclusions of this study:First,it briefly introduces the premise of statistical arbitrage implementation,concept,quantitative transaction theory and statistical arbitrage strategy model theory.Secondly,this paper introduces the design of statistical arbitrage process,including the selection of trading assets,the calculation of spread sequences,the design of trading signals and the performance evaluation after backtesting.It focuses on two models of statistical arbitrage,including Bollinger Band Mean Regression and Ito Timing Strategy.Thirdly,in the empirical research part,this paper takes the principal contract and sub-prime contract(RB1801 and RB1805)of rebar futures as the inter-period arbitrage trading method as an example,and analyzes the two statistical arbitrage strategies involved in detail.The data of three different frequencies of 1min,5mins,and 15 mins were selected for comparative study to compare the arbitrage results at different frequencies.By applying the transaction parameters obtained from the training in the sample to the data outside the sample,the analysis results of the profit within the sample and the sample from the arbitrage success rate,yield rate,return volatility,etc.are analyzed to verify the effect of different statistical arbitrage strategies and draw the corresponding conclusions.Analysis conclusion:After empirical analysis and comparison of the two data types at various data frequencies,the following conclusions are drawn: In comparison with the frequency of horizontal data,the higher the data frequency,the greater the number of transactions and the stronger profitability,but the lower arbitrage success rate.And the stability in risk control is poor.Compared with the two vertical strategies,the Bollinger band strategy has higher yield and more profitability than the Ito strategy in each frequency data.However,the Ito Timing Strategy can effectively filter out some wrong trading signals and the number of transactions is less,but the success rate is higher,asset growth is more robust,and plays an important role in risk control.